For PSS to become smaller, pages have to be either unmapped or evicted from main memory by the OS.
Probably the only way to force SQLite to unmap the memory is to set a new mmap_size (possibly after closing and reopening the database connection, which will cause a boatload of IO that most certainly will overwhelm any performance gains). The OS will typically evict pages from main memory only if it needs fresh memory and none is available in free memory. Under high enough memory pressure, unix based OSses will kill the process they think is hogging the most memory (this could be a completel y unrelated process, or possible the process that has gone haywire and is requesting insane amounts of memory), repeatedly, until the pressure subsides, while overall performance declines. -----Ursprüngliche Nachricht----- Von: sqlite-users [mailto:sqlite-users-boun...@mailinglists.sqlite.org] Im Auftrag von Nick Gesendet: Donnerstag, 19. April 2018 04:07 An: sqlite-users@mailinglists.sqlite.org Betreff: [EXTERNAL] Re: [sqlite] Any operation to trigger osMunmap? So the PSS will not decrease even if the db becomes smaller after some DELETE/vacuum operations? I think it is a better way to free the mmap memory after every query automatically inside sqlite. Why not? -- Sent from: http://sqlite.1065341.n5.nabble.com/ _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users ___________________________________________ Gunter Hick | Software Engineer | Scientific Games International GmbH | Klitschgasse 2-4, A-1130 Vienna | FN 157284 a, HG Wien, DVR: 0430013 | (O) +43 1 80100 - 0 May be privileged. May be confidential. Please delete if not the addressee. _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
